This approach is similar to FriendFeed’s schemaless database framework. The key difference is in the data locality. The goal is not to build an effective schemaless database on top of a relational database, but rather to accomodate for rapidly-evolving relational schemas and reducing the difficult of migrating forward. Hypothesis In an evolving relational (SQL) database schema, we store two types of data: Data we will be querying against and data we will be displaying.
